Enriching Great Britain's National Landslide Database by searching newspaper archives
Enriching Great Britain's National Landslide Database by searching newspaper archives
复制标题
DOI：
10.1016/j.geomorph.2015.05.019
复制
发表时间：
2015-11
期刊：
影响因子：
3.9
通讯作者：
Faith E. Taylor;B. Malamud;K. Freeborough;D. Demeritt
中科院分区：
文献类型：
--
作者：
Faith E. Taylor;B. Malamud;K. Freeborough;D. Demeritt
我们对滑坡灾害和影响最大的地方的理解主要是基于我们对过去事件的了解。在这里，我们提出了一种方法，以补充现有的记录滑坡在英国通过搜索电子档案的区域报纸。在英国，英国地质调查局（BGS）负责更新和维护国家滑坡数据库（NLD）中的滑坡事件及其影响记录。NLD记录了英国16，500多起山体滑坡事件。全国民主联盟的数据来源包括实地调查、学术文章、灰色文献、新闻、公开报告以及自2012年以来的社交媒体。我们的目标是补充丰富的全国民主联盟（一）确定额外的滑坡事件，（二）作为一个额外的来源确认的事件存在于全国民主联盟和（三）增加更多的细节，现有的数据库条目。这是通过系统地搜索Nexis英国数字档案馆在英国出版的568份地区报纸来完成的。在本文中，我们构建了一个强大的布尔搜索标准，通过实验滑坡术语的四个训练期。然后，我们将此检索应用于2006年和2012年发表的所有文章。因此，在调查的两年（2006年和2012年）中，全国民主联盟增加了111个滑坡事件记录。我们还发现，我们能够获得有关滑坡影响的60-90%的滑坡事件从报纸文章中确定的信息。从报纸文章中确定的额外的滑坡的空间和时间模式大致符合现有的全国民主联盟，确认全国民主联盟是在英国滑坡的代表性样本。这种方法现在可以应用于更多的时间段和/或其他灾害，以增加数据库的丰富性，从而提高我们根据过去事件的记录预测未来事件的能力。
Our understanding of where landslide hazard and impact will be greatest is largely based on our knowledge of past events. Here, we present a method to supplement existing records of landslides in Great Britain by searching an electronic archive of regional newspapers. In Great Britain, the British Geological Survey (BGS) is responsible for updating and maintaining records of landslide events and their impacts in the National Landslide Database (NLD). The NLD contains records of more than 16,500 landslide events in Great Britain. Data sources for the NLD include field surveys, academic articles, grey literature, news, public reports and, since 2012, social media. We aim to supplement the richness of the NLD by (i) identifying additional landslide events, (ii) acting as an additional source of confirmation of events existing in the NLD and (iii) adding more detail to existing database entries. This is done by systematically searching the Nexis UK digital archive of 568 regional newspapers published in the UK. In this paper, we construct a robust Boolean search criterion by experimenting with landslide terminology for four training periods. We then apply this search to all articles published in 2006 and 2012. This resulted in the addition of 111 records of landslide events to the NLD over the 2 years investigated (2006 and 2012). We also find that we were able to obtain information about landslide impact for 60–90% of landslide events identified from newspaper articles. Spatial and temporal patterns of additional landslides identified from newspaper articles are broadly in line with those existing in the NLD, confirming that the NLD is a representative sample of landsliding in Great Britain. This method could now be applied to more time periods and/or other hazards to add richness to databases and thus improve our ability to forecast future events based on records of past events.